In the 802.11a/g/n/ac protocol, OFDM is an important modulation and demodulation technology which can greatly improve the data transmission rate. However, the OFDM is very sensitive to carrier frequency offset and timing offset. Existence of the frequency offset and the timing offset is likely to cause degraded demodulation performance of a receiver. The frequency offset is due to the carrier frequency deviation between a transmitter and a receiver. The frequency offset may destroy the orthogonality and phase rotation value of the OFDM, and results in serious degradation of demodulation performance of the receiver. Therefore, in a communication process, an OFDM training symbol is usually added to a frame header so that the receiver may estimate and eliminate the frequency offset. However, due to noise, for the initial frequency offset estimation of a frame header of a received frame, it is not ensured that the frequency offset is accurately estimated and completely eliminated, and the frequency offset value just can be controlled within a certain range. The residual frequency offset may continue to affect the demodulation performance of the receiver. Therefore, subsequent frequency offset tracking is very necessary. The timing offset is due to the crystal oscillator frequency difference between the transmitter and the receiver or Doppler frequency drift. The crystal oscillator frequency difference may cause phase drift of a sampling clock, causing the timing offset to vary over time. The timing offset is essentially inevitable for a system. The timing offset may cause subcarrier phase rotation of the OFDM, resulting in degradation of demodulation performance of the receiver. Therefore, after timing synchronization is performed on the frame header, timing offset tracking is also very necessary. In an 802.11a/g/n/ac OFDM system, the protocol stipulates that in each OFDM symbol, several pilots known by the receiver are inserted into an information subcarrier, so that a receiver may perform synchronization tracking operation. In the prior art, there are many synchronization tracking methods performed by using the pilots. However, under impact of noise, the accuracy of frequency offset value and timing offset value that are estimated by a limited quantity of pilots is quite limited, and the demodulation performance cannot be optimized. In addition, in the prior art, frequency offset value estimation and compensation are mostly performed in time domain, thus an inverse Fourier transformer is needed, and the complexity of circuit implementation is very high.